Replace ~289s of hardcoded cy.wait() in custom-import-map-dashboards tests with cy.intercept and Cypress retry-ability

[zhongnansu]

Description

The custom-import-map-dashboards Cypress test suite contains 33 hardcoded cy.wait() calls totaling 289 seconds (~5 minutes) of unconditional sleep per test run. Zero cy.intercept() / cy.wait('@alias') calls exist. This is a primary source of both test flakiness and slow CI feedback loops.

Affected files

All under cypress/integration/plugins/custom-import-map-dashboards/:

File	cy.wait() calls	Total sleep
4_documentsLayer.spec.js	13	112s
5_add_map_to_dashboard.spec.js	7	45s
2_opensearchMapLayer.spec.js	4	36s
3_add_saved_object.spec.js	3	35s
1_import_vector_map_tab.spec.js	2	25s
6_geojson_file_upload.spec.js	1	15s
0_add_saved_object.js	2	11s
7_enable_new_home_ui.spec.js	1	10s

Anti-patterns found

1. Sleep-before-visit — Every file calls cy.wait(10000-15000) in before() after addSampleData(), then again after cy.visit(). Example in 4_documentsLayer.spec.js:19,25,27: 15s + 10s + 10s before the first assertion.

2. Sleep-instead-of-assertion — In 5_add_map_to_dashboard.spec.js:29-40, every UI interaction is preceded by cy.wait(5000). The pattern cy.wait(5000).get(selector).click() appears 6 times in a row.

3. Sleep-in-a-loop — At 2_opensearchMapLayer.spec.js:40-44, a for loop runs 21 iterations of cy.wait(1000).get('canvas').trigger('dblclick'), adding 21 seconds of dead time.

Suggested fixes

1. Replace cy.wait(N) after addSampleData() with cy.intercept on the sample data API, then cy.wait('@sampleData'). Eliminates 10-15s in every before() hook.

2. Replace cy.wait(N).get(selector) with cy.get(selector, { timeout }).should('be.visible') — Cypress's built-in retry makes this both faster and more reliable. Mechanical find-and-replace for ~20 occurrences.

3. Replace the zoom loop with programmatic map control via cy.window() to call setZoom(), or at minimum remove the cy.wait(1000) inside the loop.

4. Add cy.intercept for page-load readiness after cy.visit() instead of fixed sleeps.

Impact

• Flakiness on slower CI: if the server takes longer than the hardcoded sleep, the test fails
• Wasted time on faster CI: always sleeps the full duration even when the UI is ready in 200ms
• ~289 seconds of dead time per test run that could be reduced to near-zero with proper Cypress patterns

[andrross]

Catch All Triage - 1 2